Embed Size (px)
Citation preview
Mo
Aa
b
c
d
e
a
ARR2AA
KHVISCV
1
wat[zlgi[noW
A
0d
Vaccine 29 (2011) 2411–2420
Contents lists available at ScienceDirect
Vaccine
journa l homepage: www.e lsev ier .com/ locate /vacc ine
odelling the impact of a combined varicella and zoster vaccination programmen the epidemiology of varicella zoster virus in England
lbert Jan van Hoeka,b,∗, Alessia Melegarob,c, Emelio Zaghenid, W. John Edmundsa,b,e, Nigel Gayb,c
Immunisation, Hepatitis and Blood Safety Department, Health Protection Agency, London NW9 5EQ, UKModelling and Economics Unit, Health Protection Agency, London NW9 5EQ, UKDONDENA Centre for Research on Social Dynamics, Bocconi University, Via Guglielmo Röntgen n. 1, 20136 Milan, ItalyDepartment of Demography, University of California, Berkeley, 2232 Piedmont Avenue, Berkeley, CA 94720-2120, USACentre for the Mathematical Modelling of Infectious Disease, London School of Hygiene and Tropical Medicine, London WC1E 7HT, UK
r t i c l e i n f o
rticle history:eceived 19 July 2010eceived in revised form2 December 2010ccepted 13 January 2011vailable online 28 January 2011
eywords:erpes Zosteraricella zoster
nfectious disease modelling
a b s t r a c t
This study updates previous work on modelling the incidence of varicella and Herpes Zoster (HZ) fol-lowing the introduction of childhood vaccination. The updated model includes new data on age-specificcontact patterns, as well as data on the efficacy of zoster vaccination in the elderly and allows for HZ amongvaccinees. The current study also looks at two-dose varicella childhood programmes, and assesses thecombined impact of varicella vaccination in childhood and zoster vaccination of the elderly. The resultssuggest that a two-dose schedule is likely to reduce the incidence of varicella to very low levels, providedfirst dose coverage is around 90% and second dose coverage is in excess of 70%. Single dose varicellavaccination programmes are expected to result in large numbers of breakthrough cases. Childhood vac-cination is expected to increase the incidence of zoster for more than 40 years after introduction ofthe programme, the magnitude of this increase being influenced primarily by the duration of boosting
hingleshickenpoxaccination
following exposure to the varicella zoster virus. Though this increase in zoster incidence can be partlyoffset by vaccination of the elderly, the effectiveness of this combined strategy is limited, as much ofthe increase occurs in those adults too young to be vaccinated. Childhood vaccination at intermediatelevels of coverage (70% and 60% for first and second dose coverage respectively) is expected to lead to anincrease in adult varicella. At high coverage (90% and 80% coverage) this is unlikely to be the case. These
orm
results will be used to infprogrammes.. Background
In 1995 childhood vaccination against varicella (chickenpox)as introduced in the US [1–3]. However there is an ongoing debate
bout the potential negative effect of childhood varicella vaccina-ion on the incidence of Herpes Zoster (HZ) and varicella in adults4–7]. Herpes Zoster is a reactivation of the same virus (varicellaoster virus, VZV) that causes varicella on initial infection [8]. Fol-owing primary infection the virus remains latent in the dorsal rootanglia. Reactivation of the virus is suppressed by cell-mediatedmmunity, which can be boosted by exposure to a varicella case
9,10]. With the introduction of childhood vaccination this exoge-ous boosting would be expected to decrease due to the reductionf varicella incidence, which may lead to an increase of HZ [9,11,12].ith the licensure of a vaccine to prevent HZ [13,14], any increase∗ Corresponding author at: Modelling and Economics Unit, Health Protectiongency, London NW9 5EQ, UK. Tel.: +44 2083276065.
E-mail address: [email protected] (A.J. van Hoek).
264-410X/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.oi:10.1016/j.vaccine.2011.01.037
a cost-effectiveness analysis of combined varicella and zoster vaccination
© 2011 Elsevier Ltd. All rights reserved.
in HZ following childhood vaccination could be offset (at leastin part) by vaccination of the elderly (60+ years) against HerpesZoster. There are other concerns with varicella vaccination, whichinclude the potential increase in adult varicella (which tends tobe more serious than childhood infection) that may occur follow-ing mass childhood vaccination [11,15,16], and concerns regardingthe efficacy of a single dose of the vaccine at preventing varicella[17–19]. This latter concern has lead to recommendations for two-dose policies in childhood [3,20], which has an obvious impact onthe economic attractiveness of the programme.
Several studies have explored the effectiveness and cost-effectiveness of different varicella vaccination programmes in theUK and around the world [4,21–24]. However, to our knowledgeno previous models have looked at a combined strategy of vac-cination in childhood and of the elderly. Furthermore, previous
modelling work has concentrated on single-dose varicella vaccina-tion programmes [11,12,15,16]. Many of the indirect effects (suchas an increase of adult varicella or HZ) depend on estimates of therate of transmission across age groups. Previous analyses had toassume the relevant contact rates, as little relevant quantitative
2412 A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420
Fig. 1. Flow diagram of the model structure (bold lines are vaccination flows). WT = wild type virus; VT = vaccine type; MA = maternal antibody; V = vaccinated; p = initialvaccine failure; TV = take first vaccine dose varicella; TVii = take second vaccine dose varicella; � = duration infectious period; ˛ = duration of infectious period; �(a) = reactivationr f infec� boostb ptibilc ill re
imaegac
2
2
tBn(fciosBiv
wvbivtcc
ate; � = progression rate from vaccine protected to zoster susceptible; � = force o= Change in zoster reactivation rate in varicella vaccinees; Z = probability to beecome infectious); w = waning rate first dose; wii = waning second dose; b = susceompartment although they are vaccinated against HZ. (2) In case of � > ıv people w
nformation was available at the time. However we update previousodels [9,11,25] by using UK contact patterns collected as part of
n European project (POLYMOD) [26]. In this paper we describe thestimated impact of one and two-dose varicella vaccination strate-ies, either alone or in combination with vaccination of the elderlygainst Herpes Zoster. The described epidemiological impact willonstitute the basis for an economic evaluation of these strategies
. Methods
.1. Model structure
To assess the impact of combined varicella and zoster vaccina-ion programmes, a transmission dynamic model based on that ofrisson et al. [11] was adapted. The model consists of a set of ordi-ary differential equations and incorporates realistic age-structureRAS) and age-specific mortality rates for England and Wales (Officeor National Statistics). A stable population of 48 million people isonsidered, with 621,300 individuals born (=number of live birthsn E&W in 2003) every year on the 31st of December. The mortalityf individuals in the oldest age group (95+ year old) is calculatedeparately so that the age distribution remains constant over time.oth varicella and zoster vaccination programmes are incorporated
n the model as discrete events at the end of each year, when indi-iduals age.
The model structure is illustrated in the flow diagram in Fig. 1,hich describes the natural history of VZV with and without
aricella and/or zoster vaccination. People are initially protectedy maternal antibodies, become susceptible to varicella, can be
nfected with varicella, develop disease and become immune toaricella. After a certain time (average period of natural protec-ion = 1/ı) people become susceptible to zoster. In this state, theyan either progress to zoster, or they can be boosted by an infectiousase, rendering them (temporarily) protected against develop-
tion; K = probability to be boosted after contact when you are vaccine protected;ed after contact when you are zoster susceptible; s = duration of latency (rate toity of vaccinated individuals. (1) In case of ı > ıv, people will remain in the immunemain in the immune compartment although they are vaccinated against HZ.
ing zoster. Therefore within the model infectious people can dotwo things – infect susceptible people or boost zoster suscepti-ble persons. In case of vaccination against zoster, individuals passinto the vaccine protected compartment. These people becomezoster-susceptible again over time (average period of vaccine pro-tection = 1/ıv). When natural protection is assumed to be longerthan vaccine-induced protection, vaccinated individuals are notmoved from the immune compartment into the vaccination com-partment because in that case vaccination will reduce their timeprotected.
Vaccinated individuals are tracked separately because of the dif-ference of infectiousness and severity of breakthrough cases. In themodel described by Brisson et al. [11] individuals who had beenimmunised against varicella were not able to develop zoster. In thecurrent model this possibility is allowed for, as studies suggest thatvaccinees can, in fact, develop HZ, though it appears that they maydo so at a somewhat lower rate [27–29]. In the new structure it ispossible to distinguish between HZ among vaccinees after break-through or after vaccination only, because in the latter zoster willbe caused by vaccine virus instead of wild type.
2.1.1. Varicella vaccinationVaccination with varicella vaccine is assumed to result in three
different outcomes. First, a proportion of individuals (p) suffer aninitial vaccine failure and remain susceptible. Second, a proportionof individuals who respond initially (1 − p) are protected from vari-cella infection ((1 − p) × T; in which T is the proportion of vaccineresponders who are protected). Third, a proportion of individualsrespond, but they are liable to be infected (i.e. become a break-
through case) if they are exposed. This proportion is therefore((1 − p)(1 − T)). Those in the vaccine immune class (V Immune) canlose protection over time and pass into the vaccinated susceptibleclass (V Susceptible) at a rate (w). If individuals receive a seconddose of varicella vaccine, the responders (TVii) move on to a vaccine
A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420 2413
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Sero
prev
alen
ce (%
)
Age (years)
a
b
Data
Model fit
0.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0.010
20 30 40 50 60 70 80
Inci
denc
e V
aric
ella
Age
Data
Model fit
F data;( mates
piTc
2
Tvavittt
2
augmU
ig. 2. Best-fit model (line) compared to data (points): (a) is the fit to seroprevalenceboth of the latter are from MSGP4). The resulting best-fitting force of infection esti
rotected class (VP 2nd dose) and when they lose protection passnto the vaccinated susceptible class (V Susceptible) at a rate (wii).hose who do not respond to the second dose remain in whateverompartment they were already in.
.1.2. Zoster vaccinationZoster vaccination is given irrespective of a history of varicella.
hose who are still susceptible for varicella are handled as if they areaccinated against varicella and moved to the varicella vaccinatedrm of the model, this because the zoster vaccine is a high doseersion of the varicella vaccine. Within the model it is assumed thatndividuals can experience only one episode of zoster throughoutheir life, vaccination was thus not effective among individuals inhe zoster infected or immune compartments (ZI and ZR) becausehat will mean they become susceptible to zoster again.
.2. Mixing patterns
Data on the contact patterns among individuals of different
ges were collected as part of the POLYMOD project [25] and weresed to parameterize the mixing patterns assumed in the model. Aeneral base-case Who-Acquired-Infection-from-Whom (WAIFW)atrix (ˇ-matrix) was derived from a contact matrix based on theK study population (all contacts) combined with an age-specific(years)
(b) is the fit to the varicella GP consultations and (c) is the fit to the zoster incidenceand associated 95% credibility intervals are shown in (d).
transmission parameter (q) following the methodology originallydeveloped by Wallinga et al. [30]. The ˇ-matrix influences thesteady state disease incidence, which is assumed to reflect thecumulative proportion of individuals with serological evidence ofinfection by age. Therefore information about contacts is combinedwith information about the disease incidence to find a best-fittingvalue of q. In this case the q is fitted in such way that the result-ing age-specific force of infection based on the resulting ˇ-matrixfits observed seroprevalence (<20 years of age) [31] and varicellaincidence (>20 years of age) [32]. To fit the data three different val-ues for q where estimated, for the age 0–3, 4–21 and 22+, underthe assumption of differential susceptibility to infection by age-group. Fifty thousand different contact matrices were obtained bybootstrapping the individual contact data. For each of the matri-ces the transmission parameters were fitted, and the best fittingmatrix was used as the base case scenario. In the sensitivity anal-yses a set of 1000 different ˇ-matrices were obtained by rejectionsampling. For each contact matrix 750 different sets of q wheresampled, by varying q as a percentage of the most optimal q for
that given matrix. In the rejection sampling process only contactmatrices were used with at least 1% probability based on their mostoptimal transmission parameter, this to improve the speed of theprocess. Of the obtained possible matrices a subset of 1000 matriceswhere randomly selected to use in the sensitivity analyses.
2414 A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0.018
0 10 20 30 40 50 60 70 80 90
Inci
denc
e of
Her
pes
Zost
er
Age (years)
c
d
Data
Model fit
0.00
0.05
0.10
0.15
0.20
0.25
0.30
00 to 01 02 to 03 04 to 10 11 to 16 17 to 21 21 to 24 25 to 34 35 to 44 45 to 64 65+
Forc
e of
infe
c�on
A
( Cont
2
cbzttratfia[
2
tp
2
V
Fig. 2.
.3. Reactivation rate
In the model the ˇ-matrix influences the steady state vari-ella incidence, and the incidence of Herpes Zoster is determinedy the reactivation rate. This reactivation rate is fitted to theoster incidence data given the ˇ-matrix and the assumption abouthe duration of protection acquired by boosting. In the sensi-ivity analyses for each of the 1000 iterations the reactivationate was refitted based on the different ˇ-matrix and assumptionbout the duration of protection. Age-specific zoster reactiva-ion rates �(a) (see Brisson et al. [9] for formula used) weretted by maximum likelihood to the age specific zoster incidences found in the 4th Morbidity Survey in the General Practice31].
.4. Biological parameters
For varicella the duration of the infectious period was assumedo be 7 days after a latent period of 14 days [11]. The infectiouseriod for zoster is also assumed to be 7 days.
.5. Efficacy of varicella vaccine
In this paper we reviewed the two dose schedule of vaccinearivax (Merck/SPMSD). Unfortunately only one trial is available
ge group
inued. )
investigating the efficacy of a two dose schedule [33], this trial hasseveral short-comings. Firstly the number of plaque forming unitsof the vaccine used in the trial was higher than the licensed ver-sion; 1350 plaque forming units in the licensed vaccine comparedto a minimum of 2900 in the trial. Secondly the age distributionwas wide (4–12 years), and each age group face a different force ofinfection making it hard to extrapolate to the effectiveness of vacci-nation at 1 year old. Thirdly the presented results include the years1996 and onwards, in those years widespread vaccination was inplace in the US. Due to the dramatic change in force of infectioncaused by the vaccination programme those years were droppedfrom the analysis. Because no other trial data was available we haveused this data.
Values for take (T) and waning (W) were fitted with a simplemodel by maximum likelihood (Fig. A1-1), assuming that the forceof infection (�) = 0.2 per year, and the rate of flow to zoster sus-ceptible (�) = 1/20 per year. To reduce the number of parametersdescribing vaccine efficacy we assumed that the susceptibility ofvaccinated individuals to become infected (parameter b, Fig. 1)is the same as non-vaccinated people (hence takes a value of
100%). Vaccinated people are as likely as non-vaccinated people tobecome boosted when they come into contact with a varicella case(described by parameter k – assumed value of 100%). For the sensi-tivity analysis 1000 sets of take (T) and waning (W) were obtainedby rejection sampling.
A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420 2415
0 20 40 60 80 100
020
040
060
080
010
0012
00
Years after start program
a
b
Inci
denc
e pe
r 10
0,00
0
−5 15 35 55 75 95
020
040
060
080
010
0012
00
5 25 45 65 85 105
020
040
060
080
010
0012
00
−10 10 30 50 70 90
020
040
060
080
010
0012
00
Vaccinating only elderly
Childhood vaccination − 1 dose
Childhood vaccination − 2 doses
Childhood vaccination − 2 doses + vaccinating elderly
0 20 40 60 80 100
010
020
030
040
050
060
0
Years after start program
Inci
denc
e pe
r 10
0,00
0
−5 15 35 55 75 95
010
020
030
040
050
060
0
5 25 45 65 85 105
010
020
030
040
050
060
0
−10 10 30 50 70 90
010
020
030
040
050
060
0
Vaccinating only elderly
Childhood vaccination − 1 dose
Childhood vaccination − 2 doses
Childhood vaccination − 2 doses + vaccinating elderly
F xes shb elderd e); anH
2
Hzttp
ig. 3. Model results on the incidence of varicella (a) and Herpes Zoster (b). The boy the model. The different shading represent different programmes: black boxesark grey two-dose infant vaccination (90% first dose, and 80% second dose coveragerpes Zoster 25 years after vaccine introduction is shown in (c).
.6. Efficacy of zoster vaccine
Parameters describing vaccine take and waning associated with
Z vaccine were estimated by fitting a model (Fig. A1-2) to theoster vaccine trial [14]. The vaccine efficacy is age dependent buthe available data is not detailed enough to estimate age specificake and waning rates. Therefore a previously estimated duration ofrotection was assumed and subsequently the take was fitted [34]ow the inter-quartile range the whiskers 10–90% of the range of results generatedly vaccination only (70% coverage); white 1 dose infant varicella (90% coverage);d light grey combined infant and elderly programme. The age-specific incidence of
(Appendix 1). The base-line proportion of people who are immuneand protected was based on the placebo group in the clinical trial.For the sensitivity analyses 1000 sets of take and waning combina-
tions were obtained.The probability of developing zoster after vaccination was set sothat the incidence will be lower in vaccinated people. Two param-eters describe this process � and � (Fig. 1). � describes the rate atwhich individuals in the vaccine protected class become suscep-
2416 A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420
0
200
400
600
800
1000
1200
1400
<1 1-4 5-14 15-24 25-44 45-64 65-74 75+
Zost
er in
cide
nce
per
100,
000
Age
Before vaccina�on
Vaccina�ng only elderly
Vaccina�on only children, 2 doses
Vaccina�on children (2 doses) and elderly
c
( Cont
trawtiTmaidciwwg05
itM
2
m
---
-
crbbpb
Fig. 3.
ible to boosting (or development of zoster), and � describes theeduction in the probability of developing zoster by age given thatn individual is susceptible to boosting, but has been vaccinatedith varicella vaccine. As the data are simply reports of the reduc-
ion in zoster incidence in vaccinees compared to non-vaccinees,t is not possible to identify these two parameters independently.hus � was set to be equal to 0.05 (based on Brisson’s base-case esti-ates for the duration of a boost from a wild-type infection [11]),
nd � was estimated to give the required long-term reduction in thencidence of zoster. Furthermore as these studies generally do notistinguish whether zoster cases in vaccinees are caused by the vac-ine strain or wild-type infection the rate of development of zostern these two groups was set to be equal. In the base case scenario �
as set in such way that among the vaccinees the zoster incidenceill be 10% of that of zoster before vaccination, assuming no back-
round boosting (Appendix 2). This percentage is varied betweenand 100% in the sensitivity analysis (with such distribution that
0% is below 10% and 50% above the 10%).VZV was assumed to be at endemic equilibrium prior to vaccine
ntroduction, and the model was run for 100 years after the start ofhe vaccine programme. The model was programmed in Berkeley
adonna 8.3.14.
.7. Vaccination policies
The following vaccination strategies were considered in theodel simulations:
single dose childhood programme (1 year of age)two-dose childhood programme (1 and 3 years of age)single dose adult vaccination programme against HZ (70 years ofage [35])combined programme of two varicella doses in childhood and 1dose adult vaccination against HZ.
The base-case coverage was assumed to be 90% for the first vari-ella dose, and 80% for the second varicella dose. Only those who
eceive the first dose, get a second dose (that is there is assumed toe no catch-up of unvaccinated individuals at 3 years of age). Thease-case age at which HZ vaccine is given (70 years) was basedartly on the results of an economic analysis [34], and partly on theasis of guidance from the JCVI subcommittee [35]. The base-casegroup
inued. )
coverage for the elderly vaccination is assumed to be 70%. In thesensitivity analysis the impact of vaccination coverage was investi-gated using lower coverage rates. The duration of protection due tocontact with wild type virus or zoster vaccine was elucidated look-ing to the extreme scenarios as possible in the parameterization.
3. Results
3.1. Comparison to epidemiological data, and estimation of forceof infection
Fig. 2 compares the best-fitting model fits with the age-specificseroprevalence data [31], age specific incidence of varicella GP con-sultations [32] and the age specific incidence of Herpes Zoster [32].Although the model describes infection and not GP visits 100%reporting was assumed for varicella above the age of 20 years andfor Herpes Zoster in all age groups. The estimated force of infectionwith associated 95% credibility intervals is shown in Fig. 2d. As thefigures show, a good fit was obtained.
3.2. Dynamics of varicella and zoster post vaccination
Fig. 3 shows predicted impact of alternative vaccination pro-grammes on the incidence of varicella (a) and Herpes Zoster (b)over time. The model predicts that a single dose of varicella vaccineis likely to result in substantial numbers of breakthrough varicellacases in the long-run with an incidence around 330 (190–447) per100,000 per year. On the other hand, vaccinating infants at suchhigh levels of coverage with a two dose schedule is expected toresult in a large reduction of varicella in both the short and long-term. However, there is predicted to be an increase in Herpes Zosterincidence for about 40–60 years after varicella vaccination. Theincidence of zoster is expected to increase by up to 20% (12–36%)to almost 620 per 100,000 per year. With only vaccination againstHerpes Zoster there is no reduction in the incidence of varicellaand a modest reduction in the incidence of Herpes Zoster. In caseof a combined programme there is estimated to be a large reduc-
tion of varicella but still an increase of Herpes Zoster, although to alesser extent than only childhood vaccination, a median increase of10% (5.6–25%). This is because a large increase in zoster incidenceoccurs in middle-aged adults, i.e. adults who are too young to bevaccinated by a programme targeted at the elderly (70 years) (see
A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420 2417
5 25 45 65 85 105
020
040
060
080
010
0012
00
Years after start program
a
b
Inci
denc
e pe
r 10
0,00
0
−10 10 30 50 70 90
020
040
060
080
010
0012
00
−5 15 35 55 75 95
020
040
060
080
010
0012
00
0 20 40 60 80 100
020
040
060
080
010
0012
00
Childhood − 1st dose: 100% ; Childhood − 2nd dose: 100% ; Elderly: 100%
Childhood − 1st dose: 90% ; Childhood − 2nd dose: 80% ; Elderly: 70%
Childhood − 1st dose: 80% ; Childhood − 2nd dose: 70% ; Elderly: 60%
Childhood − 1st dose: 70% ; Childhood − 2nd dose: 60% ; Elderly: 50%
5 25 45 65 85 105
010
020
030
040
050
060
0
Years after start program
Inci
denc
e pe
r 10
0,00
0
−10 10 30 50 70 90
010
020
030
040
050
060
0
−5 15 35 55 75 95
010
020
030
040
050
060
0
0 20 40 60 80 100
010
020
030
040
050
060
0
Childhood − 1st dose: 100% ; Childhood − 2nd dose: 100% ; Elderly: 100%
Childhood − 1st dose: 90% ; Childhood − 2nd dose: 80% ; Elderly: 70%
Childhood − 1st dose: 80% ; Childhood − 2nd dose: 70% ; Elderly: 60%
Childhood − 1st dose: 70% ; Childhood − 2nd dose: 60% ; Elderly: 50%
Fig. 4. Sensitivity of base-case model results to different levels of first and second dose coverage. The overall incidence of varicella and zoster is shown.
2418 A.J. van Hoek et al. / Vaccine 29 (2011) 2411–2420
Table 1Model parameters.
Model parameters Mean value (CI) Source
Biological parametersDuration of maternal protection (months) (12/ε) 6 AssumedDuration of latent period (days) (365/�) 14 [11]Duration of infectious period (days) (365/˛) 7 [11]Duration of immunity to zoster after varicella infection (years) (1/ı) 20 [11]Proportion of effective varicella contacts that boosts against zoster (z) 100% [11]Varicella vaccine efficacy parametersRate of varicella acquisition of vaccines compared to non vaccines (b) 100% AssumedProportion of temporarily protected individuals who become immune due to contact with varicella (k) 100% AssumedRate at which temporarily protected individuals become susceptible to Herpes Zoster (1/year) (�) 0.05 Duration of protection
of natural boostingRate of varicella infectiousness of vaccines compared to non-vaccinees (m) 50%Coverage varicella: first dose 90%Coverage varicella: second dose 80%Change in the reactivation rate (�) 0.052 (0.021–0.793) See Appendix 2
Percent of individuals who become temporarily protected after zoster vaccination by age at vaccination FOI65+ = 0.0179, 1/ıv = 7.5 See Appendix 1
Zoster vaccine efficacy parameters59–64 91%65–69 81%70–74 58%75–79 50%80–84 21%85+ 9%Coverage level 70%
First dose Second dose
1/yea
Ftt
aAdtestblb
tipHpdeicetdtz
tnfiat
the use of data on observed age-specific contact patterns from apopulation-based survey [26] and subsequent uncertainty in thecontact rates. The incorporation of these data, necessitates the re-estimation of zoster reactivation rates, with updated informationon the force of infection in adults due to contact with children.
0
100
200
300
400
500
600
-10 -5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Inci
denc
e Zo
ster
per
100
,000
Natural boos�ng 20 yrs - Zoster vaccine boos�ng 7.5 yrsNatural boos�ng 42 yrs - Zoster vaccine boos�ng 3.6 yrsNatural boos�ng 7.5 yrs - Zoster vaccine boos�ng 100 yrs
Percent of individuals for which vaccine fails completely (p)Percent of individuals who become temporarily protected after vaccination (T)Rate at which temporarily protected individuals become susceptible to varicella (
ig. 2c). There is considerable uncertainty with regards estimates ofhe impact of varicella vaccination on the incidence of HZ, in bothhe medium and long-term.
Fig. 4 shows the sensitivity of these results to vaccine cover-ge, using the base case strategy and base-case model parameters.s expected, the incidence of varicella is more sensitive to first-ose coverage than second dose coverage. The short- and mediumerm incidence of zoster is not sensitive to variation in infant cov-rage levels (only with very low coverage is there a difference, nothown), as breakthrough infections are assumed to be less infec-ious than natural cases. The long-run incidence of zoster is affectedy varicella first dose coverage, as vaccinees are assumed to be less
ikely to subsequently develop zoster than those who are infectedy the wild virus (Table 1).
The sensitivity of the model results towards duration of pro-ection after a contact with a varicella case or zoster vaccinations presented in Fig. 5 assuming a two dose varicella vaccinationrogramme is combined with vaccination of the elderly againstZ. Only the base case and the two extreme combinations of theossibilities are presented. As expected assumptions regarding theuration of boosting from natural infection or vaccination of thelderly have little impact on varicella incidence. However, they donfluence the expected change in the incidence of zoster post vari-ella vaccination. When there is a short natural protection and anxtremely long protection from the zoster vaccine there is expectedo be almost no increase in HZ. In the case of very long protectionue to a natural boosting and very short protection from vaccina-ion of the elderly, there will be a more dramatic increase in theoster incidence.
Changes in the long-run incidence of adult varicella (among
hose aged 15 and over) are shown in Fig. 6, for a two-dose vacci-ation strategy aimed at children. Coverage was lowered from 90%rst dose and 80% second dose to 50% and 40% coverage for the firstnd second dose respectively. It can be seen that at high levels ofwo-dose coverage the model predicts that a decrease in adult cases4%100% (93–100%) 100% (97–100%)
r) (w) 0.04 (0.067–0.015) 0.013 (0.026–0.005)
is likely, whereas at the lower level of coverage, especially below70% an increase in adult disease, with an increasing contribution ofnatural varicella in the overall varicella burden after vaccination.
4. Discussion
We present an updated analysis of the possible impact of vari-cella and/or zoster vaccination on the incidence of varicella andzoster. A number of changes have been incorporated into this modelcompared to previous work [11]. The most important of which is
Years a�er start of the program
Fig. 5. Incidence of zoster for different durations of natural boosting and vaccineprotection. From the different combinations of the two durations only the base caseand the two extremes are shown.
A.J. van Hoek et al. / Vaccine
0
50
100
150
200
250
300
350
Before vaccina�on
90%/80% 80%/70% 70%/60% 60%/50% 50%/40%
Inci
denc
e pe
r 10
0,00
0
Vaccine coverage
Overall
Breakthorugh varicella
Natural varicella
Fio9
IvpbmrTipclto
ioilms(tititcNv
brceiosicluziedq
ig. 6. The post-vaccination equilibrium (at 100 years) annual incidence of varicellan adults age 15 and above by levels of coverage (2 dose combined policy). The heightf the bar is the median of 1000 simulations, and the error bars represent 2.5 and7.5 percentiles.
mportantly, the model now also includes a two-dose schedule andaccination of the elderly to reduce the incidence of zoster (thearameters describing this being based on analyses of a double-lind placebo controlled trial of the zoster vaccine). The updatedodel also allows for vaccinees to develop zoster, though at a
educed rate compared to those experiencing natural protection.he average duration of boosting against zoster (exogenous boost-ng) following exposure to VZV remains an uncertain and influentialarameter. As can been seen in Fig. 3 the joint uncertainty in theontact rates, duration of natural protection and other epidemio-ogical parameters lead to wide confidence intervals, meaning thathe presented outcomes are more disease trends, since individualutcomes are dependent on a spectrum of parameter values.
The results of the model with regards a one-dose policy are sim-lar to those published by Brisson et al. [11,16]. That is, vaccinationf infants at achievable levels of coverage in the UK (around 90%)s likely to result in a rapid decline in incidence of varicella, fol-owed by a period of low incidence (for perhaps a decade), which
ay be followed by a series of epidemics, before the system finallyettles around a new level of incidence which is considerably loweri.e. reduced by about 75%) than in the pre-vaccine era. Break-hrough varicella makes up a substantive portion of these cases. Noncrease in adult varicella is expected with our base-case parame-ers. However, if lower coverage is achieved (<70%) than an increasen varicella among adults is expected in the long run. Our results forhe varicella only schedules are comparable with the recent publi-ations by Karhunen et al. [23], Gao et al. [24], and Brisson et al. [4].one of these publications incorporate zoster vaccination on top ofaricella vaccination in a combined schedule.
Two-dose routine vaccination at the coverage levels that maye achievable in the UK (90% and 80% for the first and second doseespectively), is expected to result in very low incidences of vari-ella among all age groups in the long-run and may even lead tolimination of the virus. However, the magnitude of this reductions dependent on our assumptions on the efficacy of the second dosef vaccine, for which there is little good data. As with the one-dosetrategy, vaccination of young children is expected to result in anncrease in zoster in the medium term. This increase in incidencean be partly attenuated by routine vaccination of the elderly. Theong-run incidence of HZ following vaccination of children is highlyncertain, as this depends on the likelihood of vaccinees developing
oster, either via the vaccine strain, or from wild-type breakthroughnfection. The data on this are scarce, and no consistent pattern hasmerged. Most, but not all, studies have suggested that the inci-ence of zoster in vaccinees is likely to be reduced [25,28], but exactuantification of this risk is difficult [28,29].29 (2011) 2411–2420 2419
As with all models the findings are reliant on the reasonable-ness of the assumptions made and the values of the parameters.We estimated the force of infection of varicella for people abovethe age of 20 based on the incidence of GP notifications. This is notideal because there is a possibility varicella was misdiagnosed asHerpes Zoster or vice versa. More data on the actual force of infec-tion in adults would be welcome since the changes in Zoster andadult varicella incidence after vaccination depends on this. A keyassumption with regards zoster epidemiology is the degree andduration of boosting that results from exposure to the virus. Weused Brisson et al.’s best fitting estimates in our base-case model(an average duration of boost of 20 years). In this estimation theduration of boosting was identical for all ages. Brisson et al. [4] haveshown that an age dependent duration of boosting can decrease therelation between varicella and zoster, leading to a smaller increaseof zoster after vaccination. By fitting models to the data from thelarge-scale clinical trial of the zoster vaccine [14], van Hoek et al.[34] estimated that the average duration of vaccine-induced pro-tection may be significantly shorter than Brisson et al.’s estimates(best fitting estimates are in the range of 3.6–100 years). The shorterthe duration of boost, the smaller the increase in zoster followingchildhood varicella vaccination (Fig. 5). It is therefore possible thatthe base-case results overestimate the post-vaccination increase inzoster. Even using Brisson et al.’s estimates of the duration of boost,our estimates of the increase of zoster following varicella vaccina-tion are lower than those of Brisson et al. [11,16]. This is becausewe re-estimated the force of infection in adults using the POLY-MOD contact survey, which resulted in a revision of our estimatesof the risk of reactivation. Although the data on the incidence ofHZ in the US are limited and contradictory, the evidence suggeststhat the incidence is probably increasing (four of the five publishedstudies reports an increase in HZ [36–40]), though whether thisis attributable to varicella vaccination, or some other factors (likeincreased use of corticosteroids) is less clear. Further surveillancedata on zoster (accompanied by good varicella coverage and inci-dence data) is clearly needed.
We used Kuter et al.’s data to re-estimate the vaccine efficacyparameters [33]. A number of these, particularly those concernedwith breakthrough varicella, are uncertain. However, since atwo-dose schedule would be expected to reduce the number ofbreakthrough cases to low levels, and it is likely that two-doseschedules will be adopted, then this uncertainty does not signif-icantly affect findings.
This study updates previous work [11,16] on the impact of vac-cination on the incidence of both varicella and zoster. The resultssuggest that a single-dose policy may result in significant numbersof breakthrough cases, a pattern which has been observed in theUS [17]. A two-dose schedule is likely to lead to a low incidenceof varicella, provided coverage is maintained at around 90% for thefirst dose. An increase in zoster incidence is still expected follow-ing varicella vaccination in childhood, and this increase can onlybe partly ameliorated by introduction of zoster vaccination in theelderly.
Acknowledgements
This research was partly funded by POLYMOD, a European Com-mission project funded within the Sixth Framework Programme,contract number: SSP22-CT-2004-502084, and partly by a grant
from the UK Department of Health Policy Research Programme(grant number: 039/0031). AM was also partially funded by ECDCGRANT/2009/002. The views expressed in the publication are thoseof the authors and not necessarily those of the Department ofHealth.
2 ccine
rMsmtt
A
t
R
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
420 A.J. van Hoek et al. / Va
Contributors: AJvH involved in coding, parameterization, modelun and writing up, AM programming the model in Berkeleyadonna and initial validation and writing up, EZ assisted on boot-
trap of the contact matrices, JE supervision, model structure andajority of the writing up, NG involved in methodology, parame-
erization, model structure and supervision. All authors agreed withhe final draft.
ppendix A. Supplementary data
Supplementary data associated with this article can be found, inhe online version, at doi:10.1016/j.vaccine.2011.01.037.
eferences
[1] Recommendations for the use of live attenuated varicella vaccine. Amer-ican Academy of Pediatrics Committee on Infectious Diseases. Pediatrics1995;95(5):791–6.
[2] Prevention of varicella. Update recommendations of the Advisory Committeeon Immunization Practices (ACIP). MMWR Recomm Rep 1999;48(RR-6):1–5.
[3] Marin M, Guris D, Chaves SS, Schmid S, Seward JF. Prevention of varicella: rec-ommendations of the Advisory Committee on Immunization Practices (ACIP).MMWR Recomm Rep 2007;56(RR-4):1–40.
[4] Brisson M, Melkonyan G, Drolet M, De Serres G, Thibault R, De Wals P. Modelingthe impact of one- and two-dose varicella vaccination on the epidemiology ofvaricella and zoster. Vaccine 2010;28:3385–97.
[5] Rimland D, Moanna A. Increasing incidence of Herpes Zoster among Veterans.Clin Infect Dis 2010;50(April (7)):1000–5.
[6] Carville KS, Riddell MA, Kelly HA. A decline in Varicella but an uncertainimpact on zoster following Varicella vaccination in Victoria, Australia. Vaccine2010;28(13):2532–8.
[7] Donahue JG, Kieke BA, Gargiullo PM, Jumaan AO, Berger NR, McCauley JS, et al.Herpes Zoster and exposure to the Varicella Zoster virus in an era of Varicellavaccination. Am J Public Health 2010;100:1116–22.
[8] Hope-Simpson RE. The nature of Herpes Zoster: a long-term study and a newhypothesis. Proc R Soc Med 1965;58:9–12.
[9] Brisson M, Gay NJ, Edmunds WJ, Andrews NJ. Exposure to varicella boostsimmunity to herpes-zoster: implications for mass vaccination against chick-enpox. Vaccine 2002;20(19–20):2500–7.
10] Thomas SL, Wheeler JG, Hall AJ. Contacts with varicella or with childrenand protection against herpes zoster in adults: a case–control study. Lancet2002;360(9334):678–82.
11] Brisson M, Edmunds WJ, Gay NJ, Law B, De Serres G. Modelling the impact ofimmunization on the epidemiology of varicella zoster virus. Epidemiol Infect2000;125(3):651–69.
12] Schuette MC, Hethcote HW. Modeling the effects of varicella vaccinationprograms on the incidence of chickenpox and shingles. Bull Math Biol1999;61:1031–64.
13] Harpaz R, Ortega-Sanchez IR, Seward JF. Prevention of herpes zoster: rec-ommendations of the Advisory Committee on Immunization Practices (ACIP).MMWR Recomm Rep 2008;57(RR-5):1–30.
14] Oxman MN, Levin MJ, Johnson GR, Schmader KE, Straus SE, Gelb LD, et al. Avaccine to prevent herpes zoster and postherpetic neuralgia in older adults. NEngl J Med 2005;352(22):2271–84.
15] Halloran ME, Cochi SL, Lieu TA, Wharton M, Fehrs L. Theoretical epidemiologicand morbidity effects of routine varicella immunization of preschool childrenin the United States. Am J Epidemiol 1994;140(July (2)):81–104.
16] Brisson M, Edmunds WJ, Gay NJ. Varicella vaccination: impact of vaccine effi-cacy on the epidemiology of VZV. J Med Virol 2003;70(Suppl. 1):S31–7.
17] Michalik DE, Steinberg SP, Larussa PS, Edwards KM, Wright PF, Arvin AM, et al.Primary vaccine failure after 1 dose of varicella vaccine in healthy children. JInfect Dis 2008;197(April (7)):944–9.
18] Brisson M, Edmunds WJ, Gay NJ, Law B, De Serres G. Analysis of varicellavaccine breakthrough rates: implications for the effectiveness of vaccinationprogrammes. Vaccine 2000;18(25):2775–8.
[
29 (2011) 2411–2420
19] Lopez AS, Guris D, Zimmerman L, Gladden L, Moore T, Haselow DT, et al. Onedose of varicella vaccine does not prevent school outbreaks: is it time for asecond dose? Pediatrics 2006;117(6):e1070–7.
20] American Academy of Pediatrics Committee on Infectious Diseases Preventionof varicella recommendations for use of varicella vaccines in children includinga recommendation for a 2-dose varicella immunization schedule. Pediatrics2007;120(July (1)):221–31.
21] Thiry N, Beutels P, Van Damme P, Van Doorslaer E. Economic evaluations of vari-cella vaccination programmes: a review of the literature. Pharmacoeconomics2003;21(1):13–38.
22] Rozenbaum MH, van Hoek AJ, Vegter S, Postma MJ. Cost-effectiveness of vari-cella vaccination programs: an update of the literature. Expert Rev Vaccines2008;7(August (6)):753–82.
23] Karhunen M, Leino T, Salo H, Davidkin I, Kilpi T, Auranen K. Modellingthe impact of varicella vaccination on varicella and zoster. Epidemiol Infect2010;138(4):469–81.
24] Gao Z, Gidding HF, Wood JG, MacIntyre. Modeling the impact of one-dose vs.two-dose vaccination regimens on the epidemiology of varicella zoster virusin Australia. Epidemiol Infect 2010;138(4):457–68.
25] Brisson M, Edmunds WJ. Varicella vaccination in England and Wales: cost-utility analysis. Arch Dis Child 2003;88(10):862–9.
26] Mossong J, Hens N, Jit M, Beutels P, Auranen K, Mikolajczyk R, et al. Socialcontacts and mixing patterns relevant to the spread of infectious diseases. PloSMed 2008;5(3):e74.
27] Gershon AA, LaRussa P, Steinberg S, Mervish N, Lo SH, Meier P. The protectiveeffect of immunologic boosting against zoster: an analysis in leukemic chil-dren who were vaccinated against chickenpox. J Infect Dis 1996;173(February(2)):450–3.
28] Gershon AA, Takahashi, Seward J. Varicella vaccine. In: Plotkin SA, OrensteinWA, editors. Vaccines. 4th ed. Philadelphia, PA: Saunders; 2004.
29] Hambleton S, Steinberg SP, LaRussa PS, Shapiro ED, Gershon AA. Risk ofHerpes Zoster in adults immunized with varicella vaccine. J Infect Dis2008;197:S196–9.
30] Wallinga J, Teunis P, Kretzschmar M. Using data on social contacts to estimateage-specific transmission parameters for respiratory-spread infectious agents.Am J Epidemiol 2006;164(10):936–44.
31] Nardone A, de Ory F, Carton M, Cohen D, van Damme P, Davidkin I, et al. Thecomparative sero-epidemiology of varicella zoster virus in 11 countries in theEuropean region. Vaccine 2007;25(November (45)):7866–72.
32] McCormick A, Fleming DM, Charlton J. Morbidity statistics from General Prac-tice, fourth national study 1991–1992: MSGP4. Series MB5:3. London: HMSO;1995. p. 1–36.
33] Kuter B, Matthews H, Shinefield H, Black S, Dennehy P, Watson B, et al. Ten yearfollow-up of healthy children who received one or two injections of varicellavaccine. Pedietr Infect Dis J 2004;23(February (2)):132–7.
34] van Hoek AJ, Gay N, Opstelten W, Edmunds WJ. Estimating the cost-effectiveness of vaccination against Herpes Zoster in England and Wales.Vaccine 2009;27(9):1454–67.
35] Joint Committee on Vaccination and Immunisation (JCVI). JCVI short state-ment on Herpes zoster vaccines. London: Joint Committee on Vaccination andImmunisation; 2010.
36] Yawn BP, Saddier P, Wollan PC, St Sauver JL, Kurland MJ, Sy LS. A population-based study of the incidence and complication rates of herpes zosterbefore zoster vaccine introduction. Mayo Clin Proc 2007;82(November (11)):1341–9.
37] Mullooly JP, Riedlinger K, Chun C, Weinmann S, Houston H. Incidence of herpeszoster 1997–2002. Epidemiol Infect 2005;133:245–53.
38] Yih WK, Brooks DR, Lett SM, Jumaan AO, Zhang Z, Clements KM, et al. Theincidence of varicella and herpes zoster in Massachusetts as measured by theBehavioral Risk Factor Surveillance System (BRFSS) during a period of increas-ing varicella vaccine coverage, 1998–2003. BMC Public Health 2005;5(June):68.
39] Jumaan AO, Yu O, Jackson LA, Bohlke K, Galil K, Seward JF. Incidence ofherpes zoster, before and after varicella-vaccination-associated decreases
in the incidence of varicella, 1992–2002. J Infect Dis 2005;191(June (12)):2002–7.40] Reynolds MA, Chaves SS, Harpaz R, Lopez AS, Seward JF. The impactof the varicella vaccination program on herpes zoster epidemiology inthe United States: a review. J Infect Dis 2008;197(March (Suppl. 2)):S224–7.
